SPECULATIONS ON STORY PUNS
Kim Binsted and Graeme Ritchie
Department of Articial Intelligence
University of Edinburgh
Scotland EH1 1HN
ABSTRACT
There is a class of joke which consists of an anecdote, which is sometimes quite long and often
has no inherently humorous content, followed by a
punchline which is a distorted form of some wellknown phrase, proverb or quotation. Usually the
punchline purports to summarise or draw a moral
from the preceding story. This genre has some
unusual aspects, from the viewpoint of conventional claims about the attributes of jokes. These
jokes also have certain structural or formal regularities which suggest that they might be amenable to computer generation. We outline how this
might be done, by decomposing the construction
of such story puns into a sequence of stages; some
of these are clearly manageable, others are less
straightforward. We also make some observations
about where such an endeavour would t within
the broader eld of humour research.
1
INTRODUCTION
A particular class of joke, which we will call the
story pun, has some unusual features from the
point of view of a formal analysis of the internal
semantic/pragmantic structure of jokes. At some
level of abstraction, the regularities underlying
story puns can be seen as analogous to those underlying certain classes of punning riddle, in that
a supercial similarity between linguistic forms is
the central mechanism.
In past work [BR94, BR97], we have shown that
punning riddles can be generated by computational rules. Here, we sketch an architecture for
a system which could generate (some simple varieties of) story puns. We also suggest that genetic
algorithms might be used to generate the kind of
highly constrained story required for this type of
joke.
This research has been partially funded by a PhD studentship from the Natural Science and Engineering Council of Canada.
Finally, we make some observations about what
such a system would demonstrate about the
nature of humour.
2
STORY PUNS
There is a particular genre of joke which is quite
common within British society, and possibly elsewhere. It consists of a story which need not itself have any humorous content (and which may
be quite long), concluding with a single punchline which is usually a summing up of the content
or moral of the story, although it may also be a
suitable last line of the narrative or an utterance
attributed to a character in the story. What is
peculiar to this genre is that the punchline is always a distorted form of some well-known saying,
usually a proverb but sometimes a famous quotation. For example, there is a well-established
proverb in the English language \People who live
in glass houses shouldn't throw stones" (meaning
that those who are vulnerable to a particular form
of criticism should not themselves make that criticism of others). This has been used as the basis
of the following joke:
Once upon a time, many years ago, there
was a chieftain in a remote tropical village
who owned an old and battered throne of
which he was very fond. One day, a visiting dignitary gave him a brand new and
ornate throne, which the chieftain had to
adopt immediately out of politeness. However, he could not bear to part with the
old throne which had served him so well,
so he stowed it away in the roof area of
his grass hut, in case it should be useful
in the future. Unfortunately the interior
structure of his hut was too imsy to support the weight of the large object, and it
crashed through the grass ceiling, falling
on the chieftain and killing him.
The moral is that people who live in grass
houses shouldn't stow thrones.
Many examples of these story puns1 have been
constructed ([MN91] contains around 150), although few survive into common circulation. It
is possible to use the construction of such jokes
as a game or pastime, in which one person stipulates the original (undistorted) maxim, and another person has to construct the story pun (this
was what occurred in the radio show reported in
[MN91]).
The structure of such a joke is interesting, as
various attributes of the main part of the story
and the punchline (and the relationship between
them) are dierent from those which are often
posited as central to more conventional jokes. In
particular:
Neither story nor punchline creates ambiguity.
The punchline does not resolve any ambiguity.
There is no incongruity within the story.
The punchline is always a suitable (congruous) ending.
The punchline does not (apparently) violate
any expectations set up by the story.
These jokes have some other noteworthy features:
The meaning of the original, undistorted proverb plays no role in the joke at all { it is
necessary only that the wording of the proverb be suciently familiar to the listener that
it is evoked by the the distorted form. This
is in contrast to jokes typied by the muchcited \One more drink and I'll be under the
host"(Dorothy Parker), where one factor in
the humorous eect is the meaning of the related phrase \under the table".
If the hearer is completely unaware of the original version of the punchline, there is no humorous eect.
If a punchline is used whose meaning is virtually identical to the distorted proverb, but
which uses dierent wording, there is no humorous eect. This is little more than a statement that these are a subclass of puns, since
puns inherently rely on their wording.
1 One of the current authors prefers the classication
\shaggy dog story" for this genre, while the other author believes that shaggy dog stories consist of long anticlimactic stories rather than complex puns.
It is essential for the humorous eect that the
maxim be distorted.2 A story in which a character living in a glass house came to grief as
a result of throwing stones, concluding with
the ungarbled proverb about such situations,
would not have the same humorous eect.
As with many other forms of pun, the listener's
reaction to a story pun is often to groan rather
than to laugh, which emphasises that the main
story portion is essentially setting up a context in
which a pun can be made. In this way, story puns
are comparable to real-life wit where someone
achieves a humorous eect by describing some
(actual) event or situation in a garbled form of
a well-known maxim.
3
SOME ASSUMPTIONS
In tackling a topic as broad and as deep as humour, or even just verbally expressed humour,
there are a number of dierent methodological
strategies that could be adopted.
One possibility (perhaps exemplied by the
General Theory of Verbal Humour [Att94]) is to
devise a very general theory which attempts to
encompass all possible phenomena in the area
under consideration. Particular studies of subspecies of humour are then carried out within
this overall framework, guided by its form and
its principles. This is logically a very sound approach, avoiding the temptation to posit ad hoc
devices, and integrating all work under one theory. Its drawback, in the current state of our
understanding of humour, is that such a theory
may be over-general to the point of vagueness,
or (if more specic) may contain arbitrary details included prior to the evidence having been
gathered. This sort of approach could be loosely
labelled as top-down.
Alternatively, one could adopt a more bottomup approach, in which detailed studies are
made of particular humorous phenomena, using
whatever theoretical constructs appear appropriate for that domain. This has the advantage that
research can be fairly concrete, to the extent of
even leading to testable computer programs, and
hence quite ne details can be examined. The
2 Another genre might be possible in which the punchline is an actual undistorted maxim and humour is produced by the indirect or unusual way in which the proverb
applies to this particular story. However, these would not
be puns, and would be dierent from the type of joke we
are considering here.
drawback, of course, is that the analyses may exist in a vacuum, unconnected to other phenomena, and the mechanisms devised may be overspecic.
In reality, some blend of these two extremes is
necessary, and is normally used. One cannot devise a general theory (top-down) without at least
keeping an eye on the data, and one cannot work
on particular data (bottom-up) without assuming (perhaps implicitly) at least some theoretical
basis. The work presented here, although sketchy
and speculative, tends towards the bottom-up
end of the scale. We believe that, in the eld of
computational humour, it is important that some
research eort should be put into nding phenomena which are tractable, or nearly tractable, with
the current state of the art, and pursuing these in
some depth. A more rmly top-down approach,
while valuable, might not reach actual computational implementation for many decades.
There is an analogy here with the way in which
linguistic theory has developed over past decades
and centuries. Much of the early work involved
the detailed development of grammars (or fragments of grammar) for particular languages, in
the absence of any over-arching theory of language. Although theory-development became a
prominent activity in its own right under the inuence of Chomsky, even modern generative linguistics started from concepts (e.g. phrase structure) which had been developed within much
more specic and pre-theoretical streams of work.
4
4.1
PROPOSED SYSTEM
Overview
We have devised a relatively crude computational
model of how story puns could be produced.
(This is not a model of how such jokes could be interpreted.) As we will explain in section 5 below,
some of the components are still underspecied in
this preliminary design, but we believe that the
role of each module is relatively well-dened.
Viewed as an algorithm, the model could be
summarised as follows:
Choose a maxim.
Distort this maxim.
Construct a semantic representation of the
distorted version.
Develop some constraints from this semantic structure.
Devise a story to meet these constraints.
We will use a worked example to illustrate how
this might be carried out.
4.2
Constructing the punchline
4.2.1 The method
The original (undistorted) proverb or saying
could be an input parameter for the process, as
in the \game" form of joke-construction mentioned in section 2 above. Alternatively, we could
assume that our proposed joke-generator would
have access to a set of proverbs, quotations and
other well-known sayings, and that it would initially choose one at random.
Construction of possible target punchlines from
a given maxim is not too problematic. The
punchline should be \similar" to the original in
some sense. The notion of \similarity" is essentially that which is used in punning riddles
(cf. [PG84]), where puns can be based on inexact
matches between words. Tactics that are suitable include metathesis (spoonerism) (e.g. \throw
stones" ! \stow thrones") and substitution of
a phonetically similar segment (e.g. \glass" !
\grass"). There might be a large number of
phrases or sentences which could be produced
from the same original saying, so heuristics might
have to be developed to choose the more productive or suitable ones. On the other hand, the next
step in the algorithm, analysing the punchline,
might succeed only on a few distorted maxims,
so that pre-selection becomes unnecessary; that
is, a set of possible punchlines could be passed
to the next phase, letting the analyser lter out
unviable variants.
4.2.2 Worked example
The maxim chosen for this example is \Look before you leap". Substituting similar sounding
words, for example, could produce several dierent distorted maxims, such as:
Hook before you heap
Rook before two sheep
Leak before you loop
Book before you reap
All of these could potentially work in a story
pun. We will use the last one, \Book before you
reap", for this example.
4.3
Analysing the punchline
4.3.1 The method
The punchline, as constructed, is still just a sequence of words, as the garbling process operated at a very low level, manipulating phonetic
information. For the punchline to inuence the
construction of a suitable story, there must be a
representation of its content. To produce this,
there must be a module capable of scanning a
sentence and constructing a symbolic representation of its meaning in some suitable formalism
(e.g. some type of logic).
4.3.2 Worked example
There are many dierent formalisms in use in
computational language processing. For the sake
of this example, we will use a logic-like language
in which operators representing the speech-act or
illocutionary force of the utterance take as their
arguments propositions (similar to those of predicate logic).3 It is quite possible that in developing a complete system, criteria for choosing or
devising a suitable formalism would become apparent.
The semantic content of \Book before you
reap" could be represented as:
IMPER( book V (hearer, Y, Time1) &
V (hearer, X, Time2), Time1
Time2).
reap
<
A possible simplication that could be considered would be to restrict the model to morals
stated as commands, so that any variant of the
punchline which could not be analysed as a simple
imperative (e.g. \Rook before two sheep") would
be eliminated.
4.4
Constructing the story
4.4.1 Simplications
Story generation is very hard, and there is no
prospect of having a good quality completely
computer-generated ctional story in the near future. Generating a story which is guaranteed to
be summed up by a given punchline is therefore
the hardest part of this preliminary design. In order to make some progress, we believe it is both
valid and feasible to constrain the type of story
3 This general idea is reminiscent of the procedural semantics of [Woo78].
in various ways, so that our generator will be
operating in a much smaller space of possibilities. There are various genres of story which seem
to have certain stereotypical attributes, involving
the types of characters, the types of events, the
types of goals that characters have, the situations
that characters may be in, etc. In particular,
fables (e.g. [Len67]), fairy stories (e.g. [CR80])
and Greek myths (cf. [SPT96]) all appear promising. These relatively stylised genres also have
the advantage of comparatively straightforward
narrative devices { they make little use of ashbacks, multiple viewpoints, etc.
In the program of research we are outlining
here, we would select one such genre as our area
of study. For the sake of the worked example,
we will choose the moral fable genre, since story
puns bear a certain resemblance to moral fables,
in that they build up to and justify a one-sentence
moral.
4.4.2 Representing the story
Once we have adopted a manageable genre of
story, the next issue is that of story representation. Although there is a considerable literature
on literary analysis of fables (cf. [Car85]), creative writing (e.g. [EP90]), and story generation
(e.g. [Mee76] and [Kan90]), there is no consensus
on how to represent, formally, the content of a
story. We would have to postulate some abstract
representation of the story, using some knowledge
representation formalism (e.g. a traditional logic,
or something of the \KL-ONE" family [BS85]),
based on the ideas proposed in these existing disciplines.
This representation, for generality, would have
to contain information not only about events,
situations and characters in the world being described; it would also have to represent the narrative presentation (e.g. the order in which events
are described), although our simplications (section 4.4.1 above) might render this rather trivial
for the examples to be studied.
4.4.3 Constraining the story
Most AI work on story generation has focussed on
methods which would actually construct a story
from some basic material, such as information
about the characters and their goals. An alternative approach, which we would like to explore,
is to attempt to dene declarative constraints or
preferences describing the formally represented
structure of the story. That is, using ideas from
the various disciplines mentioned in Section 4.4.2
above, we would formulate a set of heuristic rules
which, given a formally represented story (or candidate for being a story) would evaluate its merit
as a story. We do not believe that there is really
some abstract measure of \quality" that can be
used to rate real stories. However, there are some
very crude guidelines which can be used to separate those structures which barely count as stories
from those which are at least well-formed stories. The basis for this belief is that literature
on creative writing can be viewed as attempts
to do just this. For example, when a textbook
advises \Never switch viewpoints in the middle
of a scene" (e.g. [Dib, Bic93]), it is formulating
a heuristic that could be applied to a symbolic
representation of a story to make some estimate
of its competence or success. (Notice that it is
crucial that we have some formal, abstract representation of the story, as it would be completely
infeasible to apply such abstract heuristics to the
bare textual form of a story.)
A constraint-based (or \evaluative") approach
such as this has the advantage that the very heterogeneous requirements of the story pun can all
be accommodated. A story pun has certain requirements which might not arise in more conventional story generation:
4.4.4 Producing the story
Let us assume that we have managed to achieve
the various goals listed above { we have selected
a narrow, stylised genre of story, we have devised
a formal representation for the content (including presentation) of such stories, and we have developed a set of heuristics which will contribute
to the rating of how well a story meets our needs.
How then can we produce such a story? Although
we may have ways of telling how good a story is
once we have it fully represented, we need some
way of creating the structures. Clearly, the space
of possibilities is very large.
Here we introduce another major speculation.
Let us assume that our formal representation is
such that we can dene a compressed encoding
of it into a sequence (tuple) of values from some
limited range. That is, suppose we can re-code a
story-stucture into the form
< v1 ; v2 ; : : : ; v
n
>
The plot must (somehow) conform to (the
meaning of) the pre-determined nal punchline or \pseudo-moral".
Enough suitable concepts and vocabulary
items must be introduced in the course of the
story to make the nal punchline relatively
natural.
Nevertheless, for the story pun to be humorously eective, there should not be so much
preparation within the story that the punchline can be easily predicted.4
where each i is selected from some (fairly small)
set of possible atomic values (and where the full
structure can be reconstructed from this sequence
of values if needed). This might well be possible,
given the simplications we have made already.
In the narrow, stylised story genres, there are often relatively few options which go to establish
the basic parameters of the story.
We are now in a position which is exactly appropriate for the application of a technique known
as evolutionary algorithms or genetic algorithms
[Mit96]. These methods (a highly sophisticated
variant on what is sometimes called \generateand-test") have proved to be very useful in problem areas where the search space for solutions is
very large, but there are clear constraints on what
constitutes a `good' solution. Genetic algorithms
are based on a loose analogy with the way that
evolution can occur through genetic variation and
natural (Darwinian) selection. They require the
following:
All these requirements are of a very varied nature,
so some very general scheme is needed for their
representation. (We do not suggest that formalising these requirements would be trivial.)
1. a class of structures or items which we wish to
produce (e.g. abstract representations of stories), regarded as analogous to organisms in an
evolutionary setting;
It is easier to succeed with a story pun if the audience
does not know in advance the approximate shape of the
punchline; it is a sign of the skills of Muir and Norden
[MN91] that their stories amuse even when the original
(undistorted) proverb is known in advance.
2. some measure of \tness" which rates how
good an item is (e.g. our heuristics of story
quality), and hence whether the item should
be given preference in a \survival of the ttest" process.
4
v
3. a way of encoding an item as a tuple, so that
the tuples can be treated as analogous to chromosomes and can be mutated or combined in
ways which regard them simply as sequences
of symbols.
4. a way of decoding a tuple as an item, so that
tness can be assessed after any mutation or
combination takes place.
This technique works in the following way. Initially, a large number of such tuples (the pseudochromosomes) are generated at random (perhaps
within some very minimal constraints on wellformedness). Then the following cycle is repeated
many times.
1. Each item is evaluated for its \tness", usually by decoding the tuple representation
(chromosome) into the full structure and then
applying various evaluation functions. This
gives some rating of the relative \goodness"
or \tness" of the various items.
2. Some manipulation of the tuples occurs, based
on their tness, in a manner analogous to
mutation or sexual reproduction in genes. For
example, the two \ttest" tuples might be
combined (reproduction); or a single tuple
might have a small number of its elements
changed randomly (mutation). This creates
a new \population" of chromosomes.
3. The cycle then repeats.
This process can be run either for some predetermined number of cycles, or until some criterion is met (e.g. that the tness of the best item
has not improved on several successive cycles).
This computational method, although over
twenty years old in its conception [Hol75], has become feasible only recently with increasing computer power. It has worked with extraordinary effectiveness in certain areas where the problem has
a large search space, a clear notion of evaluation
of \goodness", and the possibility of a quasichromosomic encoding. (See, for example, [BN95]
for a range of applications, including scheduling,
layout, stock control, nancial planning, trac
management.) Whether it would serve the cause
of story generation as well remains to be seen.
4.4.5 Worked example
Work in this area has not advanced far enough to
give even a tentative representation of the story,
let alone in a form appropriate for use with genetic algorithms. Instead, we will informally outline some of the constraints from the genre and
from the parsed punchline, and give a story that
satises those contraints.
Our chosen genre, the moral fable, has a typical setting, plot structure, cast of characters, and
prop list. Although few of these are necessary features, they help to make the genre recognizable,
and are not overly restrictive.
Characters: All characters are either animals (e.g. fox, sheep etc) or human. If human,
the character must have one of a limited number of occupations (e.g. farmer, doctor, shwife etc). Both animals and humans have
their own simple characteristics, beliefs, goals
and relations (e.g. fox is cunning, sheep is stupid, fox wants to eat sheep, farmer wants to
protect sheep etc). These characteristics do
not change over the course of the story (i.e.
there is no character development). Any character can converse with any other character.
Characters can lie.
Objects: Each character is associated with
a small set of objects. Some objects may be
associated with more than one character (e.g
hay is both eaten by sheep and mown by farmers).
Actions: Some acts can be achieved by any
character (e.g speaking), others cannot (e.g.
only farmers can mow, and only when they
have a eld of hay).
Setting: There is a limited number of settings (e.g. the forest, a farm, the city, a house
etc). Each character is typically found in one
setting, and can venture into a few others. For
example, a fox is typically found in the forest,
can go onto a farm, but would not be found in
the city. This constrains the characters that
can interact.
Plot structure: A character attempts to do
something (consistent with its goals). It fails.
The reason for its failure, and/or advice on
how to avoid a similar failure, is summarized
in the moral.
The punchline also constrains the story, in that
the fable must justify the punchline as its moral.
We assume that the system has access to lexical
information. Our chosen punchline, \Book before
you reap", might constrain the plot structure of
the fable as follows:
A character attempts to reap something,
and fails. The reason for its failure is
that it failed to book some location before attempting to reap.
From the cast of characters, objects, settings
etc, we select those that t into this outline. Let
us assume that, in `fable world', only farmers can
reap, and that the only object that can be reaped
is hay. There are no constraints from the punchline on the location, so we choose one consistent
with farmers, hay and reaping: a eld. This gives
us the basic plot:
A farmer tries to reap hay, but fails. The
reason for his failure is that he did not
book a eld before attempting to reap.
Although we suggest that the generation of the
nal version of the story could be done with some
suitable search mechanism, such as genetic algorithms, here it is done by hand.
Farmer Joe needed some hay to feed his
sheep. He picked up his scythe and
went to the eld to cut the hay. Unfortunately, when he got there, he found
another farmer already working in the
eld. He had forgotten to make a reservation!
Moral: Book before you reap.
Note that the character has been given a name,
a few objects and characters (the sheep, the
scythe, the other farmer) not strictly necessary to
the story have been introduced, and none of the
key words in the punchline are explicitly mentioned in the story. Also, the style is relatively
sophisticated, using punctuation, adverbs, etc. A
good story generator, whether based on genetic
algorithms or not, must be able to manipulate the
basic story in ways consistent with `good' creative writing heuristics, as discussed in Section 4.4.3
above.
5
HOW FEASIBLE IS THIS?
Let us summarise what would be needed to make
this work.
Punchline construction. This initial phase of
the process seems straightforward. Given some
basic lexical information (including phonetics),
distorting a sentence into a similar-sounding sentence is not dicult.
Punchline analysis. It is completely possible,
in theory, to analyse a sentence of English into
a representation of its literal meaning. In practice, it would (in the current state of the art)
not be trivial to construct a working analyser
which would provide correct analyses of arbitrary sentences with any level of syntactic complexity, formulated from a wide vocabulary. Although we are assuming that the story domain
could be simplied, the punchlines come from a
very unconstrained source, namely existing proverbs and sayings. Not only does this widen the
vocabulary greatly, it could introduce some quite
unusual grammatical forms.
Representing the story. To do this completely would require signicant research, but
(given the use of a narrow, stylised genre), it
should be feasible to develop some (comparatively
crude) formal representation. It has to be recognised, however, that taking short-cuts here could
have a deleterious eect on the story-evaluation
issue.
Evaluating a story This is the most dicult
(and perhaps most interesting) part. It would
involve signicant research, with a synthesis of
ideas from a number of subdisciplines.
Producing the story. If the previous steps,
particularly the formulation of an evaluation
measure, are achieved, then the use of genetic
algorithms should be feasible. The only further
prerequisite is the design of a quasi-chromosomic
(tuple) encoding for story structures.
6
DISCUSSION
Let us imagine that we manage to construct a
program along the lines outlined here. What
would the import of this be for a (computational)
theory of humour?
What is noticeable about the proposed design
is that none of the stages purport to embody or
rely upon a theory of humour in any way. Each
is a non-humorous process. It is the overall effect which (we hope) will be humorous. This
is not paradoxical, as it is quite acceptable for
the whole to create an eect that is not attributable to any one of its parts. However, nothing
in the design process was driven by any theory
of what was or was not funny. Rather, it was
based on observing regularities in a process which
was deemed humorous, and attempting to model
these regularities (cf. remarks in Section 3 above
on the \bottom-up" approach). If it were successful, we would have created a model of a class of
jokes without explicit use of a theory of humour.
Hence, it is not clear what such a working program would prove or disprove. It may be an experiment, loosely speaking, but what hypothesis
does it test?
It is important to realise that a computer program (or even an algorithmic but unimplemented model) may embody theoretical claims or assumptions, even although these are not explicitly
represented and even although the system designers were not consciously aware of them. Any
design has implicit assumptions, and these may
depend upon some unarticulated theory of how
humour works. It is commonplace for novice research workers in articial intelligence (typically,
students) to build computer programs without a
clear abstract or theoretical model, and to try
(with diculty) to determine afterwards what assumptions constitute their de facto theory. We
are perhaps in that position. We have to confess
ignorance of what theory of humour has led to
the outline proposal in Section 4 above.
This may be a symptom of the very early stage
of the development of formal (or computational)
models of humour. At present, we have very little
theoretical framework on which to base further
steps into the unknown. In this situation, one
possible (and perhaps essential) rst step is to
investigate and formalise what might be termed
the major structural aspects of humorous texts.
That is, a joke genre (such as story puns) will
be based on certain (fairly gross) relationships
between formal entities. These can be seen as
necessary characteristics for a text to qualify as
a member of the genre. However, some of the
texts which meet these gross structural criteria
will not be very funny at all, whereas others will
be extremely funny. Study of why there are these
variations in funniness should take us much closer
to the essential aspects of humour, and so might
seem to be more important than the modelling
of the \major structural" aspects. However, the
formal statement of the major structural aspects
provides a foundation upon which more detailed
and subtle studies can be based. Until the undergrowth has been cleared away, it is dicult to
focus clearly on the ne detail.
A loose analogy can be made with the role of
syntax in a theory of language. Even someone
who believes that the most essential aspect of lan-
guage is its meaning is likely to concede that it
is unrealistic to attempt to develop a theory of
natural language semantics in complete isolation
from syntax. Indeed, the more clearly grammatical issues are understood, the less there will be
unwanted distractions in the study of meaning.
Although the methodological points argued
above have been stated with respect to a nonexistent program (i.e. the story pun generator
whose content we have outlined), they could
largely be illustrated with respect to a working program | the JAPE riddle-generator ([BR94,
BR97]). That program generates punning riddles
which (at their best) are of a standard comparable to those circulating amongst schoolchildren. It was designed entirely by observing the
formal regularities within such riddles, and modelling these patterns in abstract rules. Although
it could be argued that the design is inuenced
by, or illustrates, ideas about ambiguity, any relationship to a general theory of humour is very
nebulous. Nevertheless, it clears the way for more
fundamental questions to be asked, such as what
types of ambiguity enhance humorous eect.
More generally, there is the point that was
already made in Section 3 above | we have to
try out lots of small scale studies just to stimulate ideas about what the ingredients might be of
a nal theory of humour.
7
WHAT NEXT?
We have argued that:
1. Story puns form a well-dened genre of joke
with characteristics quite dierent from most
told jokes.
2. There are formal regularities in story puns
which suggest a generator could be decomposed into independent stages.
3. All of these proposed modules could be constructed, at least in some simple form, using
currently known techniques.
4. This outline design, whether implemented or
not, has no apparent links to any theory of
humour in general, but might be a useful preliminary step in formalising structural aspects
of one subclass of joke.
The next steps, for those with time, inclination
and funding, would be to try implementing such
a system and then to experiment with the various
factors involved, to see what makes one story pun
funnier than another.
References
[Att94]
Salvatore Attardo. Linguistic Theories of Humour. Mouton de Gruyter,
Berlin, 1994.
[Bic93]
J. Bickham. Scene and structure.
Writers Digest Books, Cincinatti,
Ohio, 1993.
[BN95]
J. Biethahn and V. Nissan, editors.
Evolutionary Algorithms in Management Applications. Springer, Berlin,
1995.
[BR94]
Kim Binsted and Graeme Ritchie.
An implemented model of punning
riddles. In Proceedings of the Twelfth
National Conference on Articial Intelligence (AAAI-94), Seattle, USA,
1994. Available as Research Paper
690, Department of Articial Intelligence, University of Edinburgh.
[BR97]
Kim Binsted and Graeme Ritchie.
Computational rules for generating
punning riddles. Humor, 1997. To
appear.
[BS85]
R.J. Brachman and J.G. Schmolze.
An overview of the KL-ONE knowledge representation system. Cognitive Science, 9(2):171{216, 1985.
[Car85]
Pack Carnes. Fable scholarship:
an annotated bibliography. Garland,
New York, London, 1985.
[CR80]
Patricia Conroy and Sven Rossel, editors. Tales and stories of Hans
Christian Andersen. University of
Washington Press, Seattle, 1980.
[DCT95]
Ansen Dibell, Orson Scott Card, and
Lewis Turco. How to Write a Million. Robinson Publications, London, 1995.
[Dib]
Ansen Dibell. Plot. In [DCT95].
[EP90]
George Evans and Vince Powell. Get
writing: a practical guide to creative
writing. BBC Books, London, 1990.
[GSJLW86] B.J. Grosz, K. Sparck Jones, and
B. Lynn Webber, editors. Readings in Natural Language Processing.
Morgan Kaufmann Publishers, Inc.,
San Mateo, CA, 1986.
[Hol75]
John Holland. Adaptation in Natural
and Articial Systems. University of
Michigan Press, 1975.
[Kan90] Mark Kantrowitz. Natural language
text generation in the oz interactive ction project. Technical Report
CMU-CS-90-158, School of Computer Science, Carnegie Mellon University, 1990.
[Len67]
R. T. Lenaghan, editor. Caxton's
Aesop. Harvard University Press,
Cambridge, 1967.
[Mee76]
James Meehan. The metanovel :
writing stories by computer. PhD
thesis, Yale University, Department
of Computer Science, 1976.
[Mit96]
Melanie Mitchell. An introduction to
genetic algorithms. MIT Press, Cambridge, Mass, 1996.
[MN91]
Frank Muir and Dennis Norden. The
Utterly Ultimate `My Word' Collection. Mandarin Humour Classics,
1991.
[PG84]
W.J. Pepicello and Thomas A.
Green. The Language of Riddles.
Ohio State University, 1984.
[SPT96] N.
M.
Sgouros,
G. Papakonstantinou, and P Tsanakas. A framework for plot control
in interactive story systems. In Proceedings of 13th National Conference
on Articial Intelligence (AAAI-96),
pages 162 { 167, Portland, OR, USA,
1996.
[Woo78] W.A. Woods. Semantics and quantication in natural language question
answering. In M. Rubino and M.C.
Yovits, editors, Advances in Computers, pages 2{64. Academic Press,
New York, NY, 1978. Reprinted in
[GSJLW86], pp. 205-248.